efficiency and profitability of ukrainian crop …...at the institute for economic research and...
TRANSCRIPT
At the Institute for Economic Research and Policy Consulting
Agricultural Policy Report APD/APR/01/2018
Efficiency and Profitability of Ukrainian Crop Production
Dr.Marten Graubner,
Research Associate, IAMO
Igor Ostapchuk
PhD Student, IAMO
Kiew, December 2017
About the Project “German-Ukrainian Agricultural Policy Dialogue” (APD)
The project German-Ukrainian Agricultural Policy Dialogue (APD) started 2006 and is
supported up to 2018 by the Federal Ministry of Food and Agriculture of Germany
(BMEL). On behalf of BMEL, it is carried out by the mandatary, GFA Consulting Group
GmbH, and a working group consisting of IAK AGRAR CONSULTING GmbH (IAK),
Leibniz-Institut für Agrarentwicklung in Transformationsökonomien (IAMO) and AFC
Consultants International GmbH. Project executing organization is the Institute of
Economic Research and Policy Consulting in Kyiv. The APD cooperates with the BVVG
Bodenverwertungs- und- verwaltungs GmbH on the implementation of key components
related to the development of an effective and transparent land administration system
in Ukraine. Beneficiary of the project is the Ministry of Agrarian Policy and Food of
Ukraine.
In accordance with the principles of market economy and public regulation, taking into
account the potentials, arising from the EU-Ukraine Association Agreement, the project
aims at supporting Ukraine in the development of sustainable agriculture, efficient
processing industries and enhancing its competitiveness on the world market. With
regard to the above purpose, mainly German, but also East German and international,
especially EU experience are provided by APD when designing the agricultural policy
framework and establishing of relevant institutions in the agriculture sector of Ukraine.
www.apd-ukraine.de
Authors:
Dr.Marten Graubner [email protected]
Igor Ostapchuk [email protected]
Disclaimer
This work is published under the responsibility of the German-Ukrainian Agricultural
Policy Dialogue (APD). Any opinions and findings, conclusions, suggestions or
recommendations expressed herein are those of the authors and do not necessarily
reflect the views of APD.
© 2017 German-Ukrainian Agricultural Policy Dialogue All rights reserved.
Content
List of abbreviations .................................................................................................................... 4
1. Introduction ....................................................................................................... 5
1.1. Motivation and statement of the research question ............................................ 5
1.2. Structure and development of crop production in Ukraine ................................. 6
2. Data and methods ................................................................................................................ 14
2.1. Dataset ...................................................................................................................... 14
2.2. Data envelopment analysis (DEA) ........................................................................ 16
2.3. Regression analysis ................................................................................................. 17
2.4. Treatment effect analysis ........................................................................................ 19
3. Results .................................................................................................................................... 20
3.1. Development of technical efficiency and total factor productivity .................... 20
3.2. Identification of important determinants of efficiency and productivity ......... 24
3.3. Comparing of differences between farms of different profitability levels
with similar structural characteristics .................................................................... 34
3.4. Summary and discussion of the results ................................................................ 38
4. Concluding remarks ............................................................................................................. 41
References .................................................................................................................... 42
Appendix A. Summary statistics of variables for DEA model, 2008-2013 ......................... 45
Appendix B. Summary statistics of variables for DEA model for individual years,
2008-2013 ................................................................................................................................... 47
Appendix C. Total factor productivity change, 2008-2013 .................................................. 50
4
List of abbreviations
AP – animal production
BP – balanced panel
CP – crop production
CZ – climatic zone
Coef. – coefficient
DEA – data envelopment analysis
Ha- hectares
Mln – million
TE – technical efficiency
TFP – total factor productivity
TP – total production
Tsd – thousands
UAH – Ukrainian hryvnya
UP – unbalanced panel
5
1. Introduction
1.1. Motivation and statement of the research question
Because of large areas with favorable soil qualities, Ukraine features some of the prime
locations for crop production in the world. Agricultural production from Ukrainian farms
thus can play an important role to provide food for an increasing world population and,
on the other hand, can represent an important income source for Ukrainian rural
population. In fact, agriculture accounts for a significant share in gross domestic
product and export revenues. For instance, Ukraine is the fourth largest exporter of
maize worldwide in 2013 (by holding this rank since 2005) and the tenth largest
producer of wheat over the years 2011 to 2014 (FAOSTAT, 2016).
While crop production worldwide increases steadily over the last decades, Ukrainian
agriculture (as other former socialist countries) had to sustain a harsh drop in absolute
and relative production after the collapse of the Soviet Union. Only recently, the
production value of crop production (in constant prices of 2010) recovered and
exceeded the level of 1991. For most crop products, for instance wheat and maize, the
worldwide trend of increasing production is simultaneously accompanied by a reduction
(wheat) or under-proportional expansion (maize) of the harvested area (FAOSTAT,
2016), which highlights a significant increase in productivity, e.g., in terms of yields per
hectare (see Figure 1). We also obtain this trend for Ukraine, in the case of maize even
more pronounced. Despite the presumed favorable natural conditions, however,
Ukrainian average yields only recently exceeded world averages. The question therefore
arises whether Ukrainian farms exploit the yield potential of their land or if there is a
considerable yield gap that, given appropriate management, can cause significant
production growth.
In fact, the discussion of yield gaps has attracted a lot of attention over the past two
and a half decades. Per definition, yield gaps are differences between yield potential
and the average farmers’ yields for a given region and growing season, where the yield
potential refers to the yield a crop would reach under optimal conditions, i.e., without
water, nutrients, pests, or diseases stress (Lobell et al., 2009). Commonly this yield gap
is larger for developing and transition countries than for developed countries (Neumann
et al., 2010). The yield gap is smaller if the production system relies on irrigation rather
than on (uncertain and uncontrollable) rain-fed water supply (Schierhorn, et al. 2014).
These aspects are relevant for Ukraine as well as Central Eurasia - a region that
reportedly has the largest yield gaps worldwide (Neumann et al., 2010). The definition
of the yield gap, however, focuses on the natural conditions of crop production but
often or mostly ignore the question whether the maximum possible is also economically
optimal.
6
Often the argument is made that given the existence of the (seemingly considerable)
yield gap, Ukrainian farms might be able to substantially increase total crop production.
Whether this claim is justified depends however, e.g., on the economic and political
environment of Ukrainian farmers. This paper will shed some light into these issues by
investigating the productivity and efficiency of Ukrainian farms and their determinants.
Moreover, the question is tackled whether farms have a rational choice to improve
productivity given adverse political and economic conditions.
Figure 1. Yield development of major crops in Ukraine (UA) and Worldwide (W), t/ha
Source: Own calculation based on data from FAOSTAT (2016
1.2. Structure and development of crop production in Ukraine
The agricultural sector plays an important role for the Ukrainian economy. In 2013,
agriculture accounts for 8.8% of gross domestic product (GDP), 26.9% of export
revenues (excluding revenues from services), and employs 8.1% of the working
population (SSSU, 2014a; SSSU, 2014b). Besides the ongoing geopolitical frictions, the
current situation of the agricultural sector, however, is still affected by the transition
from a planned to a free market economy. Following the collapse of the Soviet Union
and Ukraine’s independence, Ukrainian farms were confronted with new and difficult
internal and external conditions including an underdeveloped banking system, the
strong reduction of subsidies paid by the government, the disconnection of former
supply chains, and the need to integrate into global markets. The farms themselves
were mostly collective enterprises with high debts and no or vague ownership of
production factors. Under these circumstances, agricultural production value
dramatically declined by 64% from 1991 to 1999. Only after new regulations for
0
1
2
3
4
5
6
7
1991 1993 1995 1997 1999 2001 2003 2005 2007 2009 2011 2013 2015
Wheat (UA) Wheat (W) Maize (UA)
Maize (W) Sunflower (UA) Sunflower (W)
7
agricultural taxation were introduced in 1999, and the improved access to capital and
infrastructure together with trade liberalizations and the farms organizational
transformation took effect (Osborne and Trueblood, 2002; Matyukha et al., 2015), the
negative trend has been reversed. However, the total agricultural production value in
2014 was still lower relative to 1991, which is mainly caused by the low level of
production value from animal production (see Figure 2). While crop production exceeds
the 1991 level in 2013 and 2014, animal production barely reaches the level of 50% of
the base year (1991) in 2014.
Figure 2. Agricultural production value, 1991 = 100%
(calculated based on constant prices of 2010)
Source: SSSU, 2016
The relative better performance of crop production underlines the favorable conditions
for this farming orientation and we are going to detail some major factors and
developments in the following.
According to official statistics, 41.5 mln hectares or approximately 71% of the total area
in Ukraine are agricultural land. Individuals or households use one third of this land
while about half of the agricultural land is operated by commercially oriented farms,
namely 20.4 mln hectares (SSSU, 2014a). The average farm size was about 2120 ha in
2013. The farm size distribution is shown in Figure 3. The comparison of the two
pictured years (2008 and 2013) indicates a slight increase in total land use by small
farms with less than 100 ha while all other groups decline in total land use. This decline
is caused by the rapid increase of land farmed by agroholdings as illustrated on the
right hand side of Figure 3. With 5731 ha per holding subsidiary, these organizations
are significantly larger than the average farm.
0%
20%
40%
60%
80%
100%
120%
19
91
19
92
19
93
19
94
19
95
19
96
19
97
19
98
19
99
20
00
20
01
20
02
20
03
20
04
20
05
20
06
20
07
20
08
20
09
20
10
20
11
20
12
20
13
20
14
Total production Crop production Animal production
8
Figure 3. Distribution of farms by the sizes of land area in use
Source: AgriSurvey, 2013
Because the total number of farms in Ukraine is decreasing, the average farm size
increases overall. However, the decline in farm numbers differs across farm types (see
Figure 4). For instance, the number of private enterprises declined by 8%, peasant
farms1 by 9%, agricultural cooperatives by 51%, and other types of enterprises by
55%, while the number of business partnerships slightly increased (+3%).
Figure 4. Number of active farms in Ukrainian agriculture
Source: SSSU, multiple years
1 This farm group mainly represents small market players with an average land area of about 100 ha but there are also large peasant farms that operate more than 20 tsd ha. In general, farms in this group
feature a dynamic increase in farm size from 92 ha in 2006 to 119 ha in 2014 (i.e., an increase of almost 30%).
0%
5%
10%
15%
20%
25%
30%
up to 100 ha 101-500 501-1000 1001-2500 2501-5000 5001-10000 10001+ Holdings
2008 2013
37000
38000
39000
40000
41000
42000
43000
44000
0
1000
2000
3000
4000
5000
6000
7000
8000
9000
2006 2007 2008 2009 2010 2011 2012 2013 2014
Peasant farms
State farms
Business partnerships
Private enterprises
Agricultural cooperatives
Other types
Peasant farms (right axis)
9
Driven by the export oriented production – more than half of the produced cereals were
exported in 2013/14 – the crop mix considerably changed over the last two decades. In
2013, the six major crops (wheat, barley, corn, sunflower, soya, and rapeseed)
accounted for 92% of the sowing area and 87% of revenues (see Figure 5 and 6). It is
noteworthy that some of these “cash crops” were only marginally important in 1990.
For instance, the share of corn increased from 3% to 20% of the total sowing area
while sunflower (from 5% to 22%), soya (from 0.3% to 9%), or rapeseed (from 0.3%
to 5%) show similar developments (see Figure 5). These crops have substituted fodder
and some niche crops, which were intended mainly for the domestic market.
Figure 5. Structure of sowing areas
Source: SSSU, multiple years
* without Crimea and occupied territories of Donetsk and Lugansk regions
Accordingly, the sales structure (see Figure 6) also features an increase in the share of
corn (from 12% in 2008 to 28% in 2013), sunflower (from 12% in 2008 to 19% in
2013) and soya (from 3% in 2008 to 7% in 2013).
0%
20%
40%
60%
80%
100%
1990 1995 2000 2005 2010 2011 2012 2013 2014*
Wheat Barley
Corn Other grains
Sunflower Soya
Rapeseed Sugar beets
Corn for silage and green fodder Other crops
10
Figure 6. Revenues from the sales of crops by farms of Ukraine, mln UAH
(in constant prices of 2008)
Source: own calculations based on SSSU (multiple years)
As initially illustrated, there is an increase in factor productivity of the input land (see
Figure 1), but also a considerable rise in labor productivity from 2000 to 2014 (about 8
times) as shown by Figure 7. Structural changes (e.g., the decline in animal production,
technological change) caused here a reduction of workforce by 79% (2014) compared
to the base year (2000).
Figure 7. Number of employees and labor productivity, 2000 = 100%
Source: SSSU, multiple years
83%
84%
85%
86%
87%
88%
0
10000
20000
30000
40000
50000
60000
2008 2009 2010 2011 2012 2013 Sh
are
of
"ca
sh
" c
rop
s,
%
Re
ve
nu
es,
mln
UA
H
Wheat Barley Corn
Sunflower Soya Rapeseed
Others Share of "cash" crops
0%
100%
200%
300%
400%
500%
600%
700%
800%
900%
0%
20%
40%
60%
80%
100%
120%
2000 2005 2010 2011 2012 2013 2014
Lab
or
pro
du
ctiv
ity
Nu
mb
er
of
em
plo
yee
s
Numer of employees Labor productivity
11
Relating to the cost level (Figure 8), we see that labor was substituted by capital
causing an intensification, which bolstered the yield development. In the period 2008-
2011 farms spent about two thousand UAH per hectare and year in average (inflation
adjusted), while the costs per hectare rose to 2.6 thousand UAH in 2012-2013. The
major components of this intensification were technological improvements (e.g., better
machinery and techniques reflected in the doubling of depreciation costs) and some
material inputs (see Table 1). The main source for increasing capital input were
comparable high profits in the years 2008-2011, but price developments in world
markets caused lower profit margins in 2012-2013, which sparks concerns regarding
the continuation of this trend.2 However, Ukrainian farms produced a record 62.2 mln
tons of cereals in 2013.
Figure 8. Total costs and revenues per hectare in crop production by farms,
(inflation adjusted)
Source: own calculations
Table 1. Production costs in CP (deflated), UAH/ha
2008 2009 2010 2011 2012 2013
Total costs 2077 1962 2027 2137 2577 2590
Labor 213 192 185 166 157 148 Social costs 51 58 66 54 83 55
Mate
rial co
sts
total 1410 1239 1278 1427 1763 1823
seeds 257 239 216 222 285 310 other agricultural products 13 13 12 13 31 17 fertilizers 366 295 296 330 359 393 oil products 313 237 253 291 409 400 electricity 21 23 25 24 24 26 fuels 20 18 16 22 36 38 spare parts 121 126 139 160 185 175
2 According to FAO, cereals price index dropped by 4.9 points in 2012 and by 16.8 points in 2013, while
for oils by 30.6 and 30.9 points respectively.
0
500
1000
1500
2000
2500
3000
2008 2009 2010 2011 2012 2013Production costs Revenue
12
3rd party services 298 289 322 366 433 466
Depreciations 117 191 194 197 240 224 O
ther
cost
s total 286 281 303 293 334 339
land rent 174 175 175 155 152 151 property rent 3 2 2 2 2 2 other 110 105 126 137 180 186
Source: own performance
As mentioned earlier, one emerging phenomena over the last two decades in Ukrainian
agriculture is the increasing importance of agroholdings, which can be defined as large
vertically and/or horizontally integrated farm enterprises. After the turn of the
millennium these organizations developed rapidly. Today, agroholdings manage almost
30% of the land in Ukraine (Figure 9) and they account for a large share of agricultural
production. In 2014, 20% of all crop products (19% in 2013) were grown or processed
by agroholdings. Relative to the average non-holding farm size in Ukraine with 1682 ha,
agroholdings are significantly larger in terms of land endowment (5731 ha per farm or
holding). Because of this size, these farms might be able to utilize economies of scale
and size and be better equipped for investments in modern technologies and
infrastructure.
Figure 9. Land bank of agroholdings in Ukraine
Source: The Largest Agroholdings of Ukraine 2015, AgriSurvey
1.702.73 3.09
4.005.10
5.60 6.04 5.85 5.60
0
5
10
15
20
25
30
2007 2008 2009 2010 2011 2012 2013 2014 2015
0
1
2
3
4
5
6
7
landbank of agroholdings, mln.ha % of land in use of agricultural enterprises
mln. ha %
13
Crop production
Households Independent farms Agroholdings
Figure 10. Share of agroholdings in crop production in 2014
Source: AgriSurvey, 2015
19.6%
39.7%
40.6%
14
2. Data and methods
To investigate whether and to what extent a potential yield gap is rooted in farm-intern
economic conditions, we conduct an efficiency analysis and regress upon the
determinants of the farms’ efficiency or inefficiency. We also approach the question if
efficiency is always accompanied with or a precondition for high profitability. The
rationale behind this procedure is as follows: if Ukrainian farms are highly efficient,
there is less room for inter-farm improvements and the perceived yield gap is the
difference between the fictive optimal yield and the actual yield of best practice. If
inefficiencies are observed, however, their sources might help to understand what hold
farms back from achieving higher productivity. Subsequently, we present the dataset
and briefly describe the methods used.
In general, this analysis follows the study by Balmann et al. (2013) but we use an
extended dataset, differentiate for production regions (climatic zones), and apply
additional methods (e.g., truncated regression and treatment effect analysis as
described below). Most of the results by Balmann et al. (2013) are supported by our
findings while some new insights and/or more detailed observations are presented in
this paper.
2.1. Dataset
Our study is based on farm-level accounting data of Ukrainian farms provided by the
State Statistics Service of Ukraine (SSSU). These data cover the time period 2008 to
2013. The original dataset consists of 51 686 observations of Ukrainian farms with crop
production and of various legal forms and sizes.
In order to control for the price development effect, the data have been deflated using
input and output price indices on disaggregated level. To eliminate inconsistent entries
and outliers as well as to prepare the data for the efficiency analysis, a two-stage data
cleaning was conducted. In the first stage three standard deviations threshold
procedure and histogram analysis were used (see Table 2 for the estimated upper limit
of ratio indicators), while the second stage identified and removed observations with
super efficiency values above 150%. As a result of the data cleaning, 6 785
observations (13%) were excluded and the obtained dataset contained 44 901
observations for a five-year (2008 to 2013) unbalanced panel and 26 982 observations
for a balanced panel (same time period). The share of agroholdings in the total number
of farms is 9(8) % in the unbalanced (balanced) panel3. Compared to the data of
statistical yearbooks, provided by the SSSU website, the unbalanced panel covers
between 85 to 91% of arable land, employees, revenue, or costs in crop production
3 The share of holdings in the original dataset also amounts to 9% of all farms.
15
through the observation period. This illustrates the representativeness of the dataset for
Ukrainian crop farms.
Table 2. Cleaning parameters for ratio indicators
Ratio 2008 2009 2010 2011 2012 2013
(Material costs + depreciation) / CP value4
<2.1 <1.8 <1.9 <1.8 <2 <1.9
Arable land / CP value <3.1 <3 <2.8 <2.1 <2.1 <1.5 Labor in CP / CP value <0.08 <0.07 <0.07 <0.06 <0.045 <0.03 CP value / (material costs + depreciation)
<5.8 <5.5 <5.5 <5.5 <4.5 <4.5
CP value / arable land <9 <9 <9 <10 <10 <11 CP value / labor in CP <830 <900 <900 <1150 <1250 <1400
Note: lower limit is always greater than 0, while upper limit is provided in the table.
Monetary values are inflation adjusted.
Source: own performance
To track peculiarities of production technologies caused by natural conditions, we
distinguish three climatic zones (production regions), which are composed of six
agroclimatic zones in Ukraine (Bulava, 2008), including two mountain regions (Crimea
and Carpathian). We merge the Carpathian region with the 1st climatic zone (1st climatic
zone: moist, moderately warm); also, two southern agroclimatic zones (2nd climatic
zone: representing “dry and very warm” and “very dry and very warm” conditions) are
subsumed. The 3rd climatic zone consists of Crimea mountainous region and the South-
Eastern regions. Figure 11 shows the location of the three zones as used in this study.
4 Note that, e.g., the share of material and capital costs in crop production value should be less than 1 for
positive profits, to be efficient, and to be productive. However, in our analysis we leave some space for
inefficient farms (eliminating only extreme or zero values) in order to obtain a comprehensive picture of Ukrainian agriculture.
16
1st climatic zone – enough moisture, moderately warm; 2nd climatic zone – not enough moisture, warm; 3rd climatic zone – (very) dry, very warm.
Figure 11. The three Ukraine climatic zones (production regions).
Source: own presentation based on Bulava (2008)
2.2. Data envelopment analysis (DEA)
In order to analyze recent productivity developments of farms, we investigate technical
efficiency and total factor productivity based on a standard Data Envelopment Analysis
(DEA) framework. This method makes use of linear programming to construct a non-
parametric piecewise surface (frontier) over the data which allows deriving efficiency
scores relative to this frontier (Coelli et al., 2005). The model is specified as a single
output - multiple input problem, and assumes output-oriented optimization with
constant returns to scale. The analysis is carried out on the balanced as well as the
unbalanced data panel for 2008-2013 with respect to individual frontiers for each year.
One disadvantage of the method is its limitation in controlling for data (measurement)
errors and effects of differences in production conditions (e.g., land quality). Therefore,
the rigorous data cleaning was carried out before the data were processed in the DEA.
Additionally, we differentiate between three production zones with different climatic
conditions.
The DEA model includes one output and three input variables. Appendix A provides
main statistics for the listed DEA variables for the years 2008-2013 and for the group of
observations in the balanced and unbalanced data panel with respect to the
corresponding climatic zones. Descriptive statistics for individual years are presented in
Appendix B. Output is represented by crop production value, which is derived from
production sales. This value thus considers real sales prices in aggregating the farms’
production volumes. Such a measure has the benefit of reflecting possible differences in
product quality along the technical ability to produce. Among the three input variables,
two are expressed in natural values (quantities) - land in hectares and labor in
17
employees in the given year (full-time employee equivalent). The third input variable is
the sum of material costs and depreciations as an indicator of capital costs. These two
input categories are pooled because a significant share of material cost refers to
services provided by third parties. Such services do not entail purely provision of
materials, but also their application that reduces the need for own machinery,
equipment or other capital.
2.3. Regression analysis
The technical efficiency scores derived from DEA as well as crop yields (wheat, corn and
sunflower) are further regressed upon a number of variables that may contribute to the
explanation of these variables’ variation among the farms. Accordingly, model
specifications detailed in Table 3 were used. For the yield regression, we used simple
ordinary least squares (OLS) regression and to estimate the technical efficiency scores a
truncated regression model was used because the dependent variable is restricted
between 0 and 1 here.
The explanatory variables take account of several structural farm characteristics such as
size, specialization, input use intensity, etc. The regression analysis considers also
control variables (time, holding membership dummies). Technical efficiency scores are
measured with respect to the individual years’ frontiers and with regard to climatic
zones. Additional variables account for size effects in groups of “smaller” and “larger”
producers. A farm size dummy [arable land > median (dummy, by CZ)] represents
farms larger than median in terms of total arable land while a crop specific harvested
area dummy [crop harvested area > median (dummy, by CZ)] relates to farms with
larger harvested area of the analyzed crop. Because we assume a positive correlation of
knowledge and year-to-year production, we test whether continuous production of a
particular crop [experienced crop producer (dummy)] realize higher yields. The last two
variables, profit per hectare in crop production [profit in CP (1 year lag)] and VAT
reimbursements in crop production [VAT support in CP (1 year lag)], allows to track for
the influence of earnings and state support in previous year on efficiency and
productivity of the analyzed year. These estimations are based on the balanced panel
data for 2008-2013.
18
Table 3. Description of explanatory variables in regression analyses of
technical efficiency, crop yield and crop production value determinants
Variable Description Model spe-cifications*
yield Yield wheat/corn/sunflower (t/ha) DV
te (bp) Technical efficiency (balanced panel) DV
harvestedarea Harvested area of wheat/corn/sunflower (ha) +
arableland Arable land (ha) +
landrent_ha Land rent per ha (tsd UAH/ha) + +
laborcosts_ha Labor costs per ha (tsd UAH/ha) + +
seeds _ha Costs of seeds per ha (tsd UAH/ha) + +
fertilizers_ ha Fertilizers costs per ha (tsd UAH/ha) + +
3rdparty_othercosts _ha
3rd party services and other material costs per ha (tsd UAH/ha)
+
otheragproducts_ha Costs of other agricultural products used in production, per ha (tsd UAH/ha)
+
oils_ ha Costs of oils per ha (tsd UAH/ha) +
electricity_ ha Electricity costs per ha (tsd UAH/ha) +
fuels_ ha Fuel costs per ha (tsd UAH/ha) +
spareparts_ ha Costs of spare parts per ha (tsd UAH/ha) +
propertyrent _ha Costs of property rent per ha (tsd UAH/ha) +
3rdpartyservices_ha 3rd party services per ha (tsd UAH/ha) +
other_matcosts_ha Other material costs per ha (tsd UAH/ha) +
other_procosts_ha Other production costs per ha (tsd UAH/ha) +
depreciation_ha Depreciations per ha (tsd UAH/ha) + +
cropshare_arableland Share of wheat/corn/sunflower area in total arable land
+
animprod_dv Animal production dummy, if farm has animal production = 1, otherwise 0
+ +
shareap_totvalue Share of animal production in total production + +
othercpvalue_cpvalue Share of niche crops in crop production (“cash” crops: wheat, barley, corn, sunflower, soybean, rapeseed; others – niche crops)
+ +
td2009 Time dummy - 2009, year 2009 = 1, otherwise 0
+ +
td2010 Time dummy - 2010, year 2010 = 1, otherwise 0
+ +
td2011 Time dummy - 2011, year 2011 = 1, otherwise 0
+ +
td2012 Time dummy - 2012, year 2012 = 1, otherwise 0
+ +
td2013 Time dummy - 2013, year 2013 = 1, otherwise 0
+ +
holding_dummy Agroholding farm dummy, agroholding farms = + +
19
Variable Description Model spe-cifications*
1, independent farms = 0 arable land > median (dummy, by CZ)
Farm size dummy, for farms > median = 1, otherwise 0. The median values were calculated for each model separately (i.e., it depends on crop and climatic zone)
+ +
crop harvested area > median (dummy, by CZ)
Crop (wheat/corn/sunflower) harvested area dummy, for farms > median = 1, otherwise 0. The median values were calculated for each model separately (i.e., it depends on crop and climatic zone)
+
experienced crop producer (dummy)
Dummy represents farms that had produced particular crop (wheat/corn/sunflower) during the whole period of analysis: if production > 0 each year during 2008-2013 = 1, otherwise 0.
+
profit in CP (1 year lag)
Profit per hectare in crop production with 1 year lag (tsd UAH/ha)
+ +
VAT support in CP (1 year lag)
VAT reimbursement in crop production per ha with 1 year lag (tsd UAH/ha)
+ +
* DV – dependent variable, “+” – independent variable,
Note: The yield model specification is crop specific (for wheat, corn, and sunflower). Each
specification consists of variables related to the production costs of the respective crop (e.g.,
wheat yield is explained by wheat production costs), while for crop production value and
technical efficiency models, total crop production costs are used.
2.4. Treatment effect analysis
In order to explain the differences between more and less profitable crop producing
farms, we employ treatment effect analysis – a matching procedure that allows the
comparison of treated and non-treated groups by pairing their structural variables (see
works of Rubin, Holland, Robins, Wooldridge for more details). In our sample the
treated group is represented by farms with crop production profitability above the
median of the base year (2008) while 50 percent of farms with lower profitability belong
to non-treated group. Profitability is measured by the relation of profit to total costs.
The advantage of this method is bias elimination related to size and structural
differences of farms. As we choose a direct-covariate matching approach, an outcome is
calculated by comparing farms (“neighbors”) with similar structural characteristics.
These “neighbors” are determined with covariates’ weighted function, calculated for
each individual. The average treatment effect (ATE) is calculated as an average
difference of observed and potential outcomes of the nearest neighbors:
𝐴𝑇𝐸 = 𝐸(𝑦1 − 𝑦0) .
20
One requirement for reliable results under the application of this method is a large
sample size, which our dataset provides5.
3. Results
3.1. Development of technical efficiency and total factor productivity
Our results show surprisingly low technical efficiency among Ukrainian farms. This not
only indicates highly heterogeneous farm performances but also a significant potential
of improvement on the farm level. In the sample, up to 30% of the farms are
unprofitable (in each year), which provides one explanation for such low technical
efficiency and indicates a high correlation between technical efficiency and profitability.
Estimations for both the unbalanced (see Table 4) and balanced panel (see Table 5)
show similar trends: an increasing mean of the efficiency scores for the 1st and 2nd
climatic zones, while there is no or even a negative change of this indicator in the 3rd
region. Overall, efficiency is slightly higher for the balanced panel, which highlights the
effects of inefficient farms that leave the sector at some point (in the unbalanced panel)
or that new (or merged) farms run through an adjustment period and might require
some time to improve their performance. But even for the balanced panel and best year
(2012), on average farms in the 1st climatic zone could (theoretically) increase output
levels by 43% without changing the level of input.
Table 4. Technical efficiency development (unbalanced panel), 2008-2013
2008 2009 2010 2011 2012 2013
Climatic zone 1
Number of observations
1462 1347 1156 1196 1119 1090
Mean 0.367 0.412 0.407 0.393 0.482 0.429 Standard deviation 0.169 0.163 0.168 0.165 0.188 0.173
Climatic zone 2
Number of observations
2314 2248 2170 2166 2185 2193
Mean 0.417 0.409 0.404 0.419 0.450 0.462 Standard deviation 0.163 0.157 0.163 0.164 0.161 0.148
Climatic zone 3
Number of observations
3883 3970 4115 4247 4022 4018
Mean 0.380 0.407 0.382 0.376 0.358 0.379 Standard deviation 0.149 0.153 0.153 0.152 0.153 0.143
Note: means are geometric Source: own calculations 5 Estimations were done using “teffects nnmatch” command in Stata software (StataCorp., 2015).
21
Table 5. Technical efficiency development (balanced panel), 2008-2013
2008 2009 2010 2011 2012 2013
Climatic zone 1
Number of observations
560 560 560 560 560 560
Mean 0.501 0.451 0.480 0.436 0.566 0.466 Standard deviation 0.176 0.156 0.153 0.164 0.169 0.156
Climatic zone 2
Number of observations
1325 1325 1325 1325 1325 1325
Mean 0.434 0.431 0.410 0.433 0.467 0.502 Standard deviation 0.152 0.147 0.152 0.152 0.149 0.137
Climatic zone 3
Number of observations
2612 2612 2612 2612 2612 2612
Mean 0.391 0.440 0.412 0.393 0.368 0.393 Standard deviation 0.141 0.144 0.145 0.136 0.136 0.134
Note: means are geometric
Source: own calculations
The development of total factor productivity (TFP), efficiency and technological change
is summarized in Figure 12 (further results including the decomposition of the efficiency
change are presented in Appendix C). Low efficiency scores might be due to high
technological change. In fact, we observe a positive development in climatic zones 2
and 3, while a negative trend is obtained in region 1. Besides that these dynamics seem
insufficient to explain the low technical efficiency, there is also considerable variation
over the observed years. Except for climatic zone 1, the annual average efficiency
change is below one, suggesting that farms become more heterogeneous in efficiency
(i.e., the distance between best and worst farms increases). The combination of both,
technological and efficiency change, provides the measure for total factor productivity
(TFP). While farms in climatic zones 1 and 3 became more productive, the opposite is
true for farms in climatic zone 2 (based on the annual average).
Of course, one of the main drivers behind the observed dynamics relates to weather
conditions. For instance, the technical efficiency drop in climatic zone 1 in the years
2010 and 2012 are largely caused by adverse weather conditions (cf. Figure 1,
representing crop yields). Other impacts concern the political framework as illustrated
by the negative TFP change in climatic zone 2. This region is the major grain producer
and therefore it was over-proportionally affected by the introduction of trade
restrictions (export quotas in 2010 and export duties in 2011), which let prices
increasingly diverge in Ukrainian and foreign markets (Kulyk et al., 2014). Farms in
22
climatic zone 3 show positive TFP change due to positive technical change while
efficiency change was slightly negative. The latter is caused by the decrease of crop
yield accompanied by increasing costs in 2009 and increasing heterogeneity across
farms (decreasing scale efficiency) over the whole time period. Pure efficiency change is
slightly positive over the considered years though. These developments are summarized
in Table C of the Appendix and illustrated in Figure 12.
Climatic zone 1
Climatic zone 2
0.700
0.800
0.900
1.000
1.100
1.200
1.300
1.400
2008~2009 2009~2010 2010~2011 2011~2012 2012~2013
TFP change Technology change Efficiency change
0.800
0.850
0.900
0.950
1.000
1.050
1.100
1.150
2008~2009 2009~2010 2010~2011 2011~2012 2012~2013
23
Climatic zone 3
Figure 12. Average annual technical efficiency, technical and total factor
productivity changes, balanced panel, 2008-2013
Source: own calculations
0.800
0.850
0.900
0.950
1.000
1.050
1.100
1.150
1.200
1.250
2008~2009 2009~2010 2010~2011 2011~2012 2012~2013
24
3.2. Identification of important determinants of efficiency and productivity
While the previous section shows the efficiency and productivity developments, we now
proceed with the determinants behind these measures in greater detail. First, we
analyze crop yield determinants (as one of the most important productivity indicators)
before we account for efficiency determining factors directly.
Estimation of crop yield determinants
As crop yields are an important productivity indicator, it is crucial to define the pivotal
factors affecting yield and if they differ among the regions (climatic zones) and crops.
In this section, we analyze three major crops: wheat, corn, and sunflower. Figure 13
shows the shares of these crops in sowing areas across the three zones (in 2013). We
note the dominant role of wheat and sunflower for climatic zone 1, while corn is the
major crop in the central and northern parts of Ukraine (climatic zones 1 and 2). More
specifically, 24% of the area is covered by wheat and 25% by corn in climatic zone 1
while climatic zone 2 features 33% corn, 32% wheat and 28% sunflower, respectively.
The yields of these crops naturally vary across climatic zones. The highest yield level is
achieved in climatic zone 2 due to favorable temperatures, humidity, and soil qualities.
In the case of corn, however, high yields are observed in climatic zone 1 as well.
Comparing the costs and revenues in this figure, we can observe some degree of
correlation. However, due to price decreases on key commodities both on the world and
the domestic markets, in average only for climatic zone 2 revenues exceed costs in case
of grains in 2013. Average profitability figures6 on the country-level, published by State
Statistic Service of Ukraine (2014c), also indicate low financial results of grain
production (2.4% for wheat and 1.5% for corn) and better results for sunflower
(28.5%).
6 As a ratio of profit to costs
25
Wheat
Corn
Sunflower
Figure 13. Distribution of sowing areas, production costs, revenues and
yields in Ukraine in 2013
Source: SSSU (2013), own representation.
3.24
4.44
3.71
3.51
5.03
4.53
3.64
4.90
4.99
0 2 4 6
CZ3
CZ2
CZ1
Costs, tsd UAH/haRevenue, tsd UAH/ha
5.48
7.55
7.12
5.04
7.42
7.15
5.28
6.97
7.19
0 2 4 6 8
CZ3
CZ2
CZ1
2.28
2.76
2.23
5.90
6.77
4.80
4.33
5.49
4.95
0 2 4 6 8
CZ3
CZ2
CZ1
26
Note: The maps represent the shares of sowing areas covered by the specific crop in
each region. The climatic zones are highlighted with a black border. The charts
represent mean values of costs, revenues and yields.
Wheat yield determinants. As Table 7 shows, fertilizer [fertilizers] is the most
limiting production factor for wheat yield. According to our model, the additional
application of fertilizer (in terms of higher input costs) of 1 thousand UAH per hectare
would cause an increase in yield by about 1 to 1.2 tons per hectare. Contrarily, seed
costs [seeds] have no significant influence on wheat yield except of climatic zone 2.
However, even in this region the effect is comparatively small. This might highlight that
most farms use seed from the last season’s harvest. Services of third-party
organizations [3rd party services] and other material costs [other material costs] also
have a significant positive impact on wheat yield. Unfortunately, we cannot decompose
these groups further, but we should note that both may include crop protection costs
(either as a service of application or as a purchase of crop protection products), while
machinery leasing and direct production and harvesting services could be also included
in third party services.
Other groups of production costs such as land rent payments [land rent per ha], labor
[labor costs in wheat production], depreciation [depreciation], and other production
costs [other production costs], have statistically significant, positive influence on wheat
yield. Concerning the labor costs in crop production, we might note that there can be a
bias as farm managers may not track for the actual time allocated to a single crop.
Thus, they – to avoid additional transaction costs – may distribute the labor costs of a
farm arbitrarily among crops. As long as this distribution is consistent over time, the
bias might be small but, nonetheless, we cannot differentiate heterogeneous labor costs
developments among the different crops. In the case of land rent payments, our results
indicate that yield is sensitive regarding land quality, where we observe higher
dependence of yield on land quality (approximated by rental payments) in regions with
less favorable soils conditions (climatic zone 1) as well as with low precipitation and
high temperatures (climatic zone 3).
The share of wheat in total sowing areas of a farm [share of wheat in sowing area] in
climatic zones 1 and 3 show statistically significant negative effect on yield. This may
suggest that the dominant role of these crops within the crop rotation is sufficiently
high to depress yields, e.g. by soil depletion or the occurrence of crop-specific weeds,
insects, and fungus. However, this effect can also relate to intra-regional differences in
soil quality and/or climatic conditions for which we cannot control. In particular, this
seems to be plausible for climatic zone 1 where a broader crop rotation (that also
includes niche crops [share of niche crops in CP]), provides a negative yield effect;
something we also observe in region 2. A positive yield effect is obtained for farms with
27
animal production [animal production (dummy)] in climatic zones 2 and 3. This is in line
with the previous finding such that farms can, e.g., use the manure to improve the
fertilizer application practices in wheat production. However, the effect of the share of
animal production [share of AP in TP] has no statistically significant influence on yields.
We further obtain some evidence of scale effects. In terms of arable land, larger farms
[arable land > median (dummy, by CZ)] achieve higher wheat yields in climatic zone 2.
The harvested area of wheat, included into the model as a continuous variable, shows a
statistically significant positive effect on a low level in climatic zone 1. However, larger
wheat producers – in terms of harvested area separated by median level [wheat
harvested area > median (dummy, by CZ)] – achieve higher yields across all climatic
zones. The holding affiliation [holding dummy] provides a positive influence only in
climatic zone 3, a region with the smallest share of land harvested by holdings
(approximately 17% compared to 28% in climatic zone 1 and 37% in climatic zone 2).
Moreover, experienced wheat producers [experienced wheat producer (dummy)], who
harvest wheat in all considered years (i.e., from 2008 to 2013) also achieve higher
yields in climatic zones 1 and 2. This might indicate learning effects, while the negative
effect in climatic zone 3 may suggest that experienced farmers apply cost minimizing
production practices to manage climate risks. The data provide some support for this
account: In wheat production and compared to climatic zone 1, farmers in climatic zone
3 had 17% lower production costs per hectare, whereas the yield difference was only
5% (see Figure 13).
Our results also suggest that higher profit per hectare received in the previous year
[profit in CP (1 year lag)] influences wheat yields positively. These effects are
consistent across all climatic zones. We also observe the positive effect of VAT
reimbursement [VAT support in CP (1 year lag)] on wheat yield in climatic zones 1 and
2, while in climatic zone 3 it is negative.
Table 6. Parameter estimates of regression model of wheat yield
determinants, balanced panel, 2008-2013
Climatic zone 1 Climatic zone 2 Climatic zone 3
Coef. P>|t| Coef. P>|t| Coef. P>|t|
harvested area (wheat) 0.000 0.021 0.000 0.806 0.000 0.877
land rent per ha 0.340 0.000 0.161 0.007 0.392 0.000
labor costs in wheat production
1.224 0.000 0.866 0.000 1.069 0.000
seeds 0.092 0.391 0.184 0.011 -0.060 0.301
fertilizers 1.192 0.000 1.070 0.000 1.020 0.000
3rd party services 0.755 0.000 0.613 0.000 0.880 0.000
28
Climatic zone 1 Climatic zone 2 Climatic zone 3
Coef. P>|t| Coef. P>|t| Coef. P>|t|
other material costs 0.679 0.000 0.591 0.000 0.797 0.000
depreciation 0.656 0.000 0.571 0.000 0.570 0.000
other production costs 0.255 0.000 0.490 0.000 0.493 0.000
share of wheat in sowing area
-0.802 0.000 0.053 0.683 -0.251 0.001
animal production (dummy)
0.054 0.198 0.199 0.000 0.167 0.000
share of AP in TP -0.062 0.461 -0.104 0.148 -0.051 0.416
share of niche crops in CP -0.400 0.000 -0.640 0.000 0.052 0.408
holding dummy 0.001 0.986 -0.036 0.373 0.183 0.000
arable land > median (dummy, by CZ)
0.065 0.160 0.162 0.000 0.026 0.219
wheat harvested area > median (dummy, by CZ)
0.200 0.000 0.088 0.013 0.145 0.000
experienced wheat producer (dummy)
0.194 0.005 0.136 0.002 -0.097 0.000
profit in CP (1 year lag) 0.246 0.000 0.255 0.000 0.224 0.000
VAT support in CP (1 year lag)
0.267 0.041 0.222 0.000 -0.221 0.000
Constant 0.840 0.000 1.435 0.000 1.565 0.000
Nr. of observations 2628
6026
12212
Prob > F 0.000
0.000
0.000
R-squared 0.654
0.547
0.498
Adjusted R-squared 0.651
0.545
0.497
Source: own calculations
Note: To keep the table at reasonable length, the time dummies considered in the
model are not reported. Arable land median: CZ1 – 1193 ha, CZ2 – 1623 ha, CZ3 –
1891 ha. Wheat harvested area median: CZ1 – 348.5 ha, CZ2 – 358 ha, CZ3 – 493 ha.
All the monetary valued are deflated and calculated per hectare of harvested area.
Corn. In general, most of the findings obtained for wheat production hold true for corn
as well. For instance, farms seem to operate on low input levels making intensification
an option to considerably increase corn yield. This is implied by positive yield effects of
fertilizers [fertilizers], third party services [3rd party services], other material costs
[other material costs], as well as other production costs like labor [labor costs in corn
production] or depreciation [depreciation]. Furthermore, we also found evidence of
economies of size, i.e., larger farms [arable land > median (dummy, by CZ)] achieve
higher corn yields in climatic zones 1 and 2. However, positive scale effects with regard
to corn area ([harvested area (corn)] and [corn harvested area > median (dummy, by
CZ)]) are observed only in climatic zone 3. Thus, the expansion of corn production (in
29
terms of sowing area) in this region may contribute positively to corn yield, which
implies potential gains from specialization. This is not the case for the major production
region (climatic zone 2), which – together with a negative yield effect of an increasing
corn share of the crop area – potentially indicates that corn area is at its upper limit in
crop rotations in this region.
In contrast to wheat production, corn seeds [seeds] are crucial in providing a
(considerable) positive yield effect. This suggests that the appropriate selection of
adopted high-quality corn varieties (hybrids) substantially increase the yield level (e.g.
Troyer, 1996). Land quality (approximated by land rent [land rent per ha]) only has a
significant positive effect in climatic zone 1, while an increase of niche crops (e.g.,
indicating diversification) positively affects corn yield in climatic zone 3 (where the crop
rotation is especially narrow). Also animal production ([animal production (dummy)]
and [share AP in TP]), which might provide organic fertilizer, can contribute a positive
effect. The affiliation with a holding has no statistically significant influence on corn
productivity.
For all climatic zones, higher profitability in the previous year [profit in CP per ha (1
year lag)], and, for climatic zone 1 and 3, the permanent engagement in corn
production [experienced corn producer (dummy)] have positive influences on corn
yield. In case of VAT reimbursements [VAT support in CP per ha (1 year lag)], we
observe a positive (negative) effect in climatic zone 2 (3).
Table 7. Parameter estimates of regression model of corn yield determinants,
balanced panel, 2008-2013
Climatic zone 1 Climatic zone 2 Climatic zone 3
Coef. P>|t| Coef. P>|t| Coef. P>|t|
harvested area (corn) 0.000 0.823 0.000 0.304 0.000 0.046
land rent per ha 0.493 0.072 0.051 0.675 -0.008 0.947
labor costs in corn production
0.838 0.000 0.501 0.000 1.130 0.000
seeds 1.233 0.000 1.016 0.000 1.337 0.000
fertilizers 1.115 0.000 1.007 0.000 1.199 0.000
3rd party services 0.805 0.000 1.014 0.000 1.250 0.000
other material costs 0.786 0.000 0.707 0.000 0.797 0.000
depreciation 0.371 0.007 0.654 0.000 0.759 0.000
other production costs 0.685 0.000 0.603 0.000 0.663 0.000
share of corn in sowing area
0.749 0.087 -1.181 0.000 1.108 0.000
animal production (dummy)
-0.198 0.136 0.099 0.205 0.256 0.000
share of AP in TP 0.917 0.001 0.436 0.010 0.206 0.180
30
Climatic zone 1 Climatic zone 2 Climatic zone 3
Coef. P>|t| Coef. P>|t| Coef. P>|t|
share of niche crops in CP -0.414 0.121 -0.717 0.000 0.366 0.053
holding dummy 0.096 0.564 -0.197 0.028 0.014 0.864
arable land > median (dummy, by CZ)
0.445 0.000 0.162 0.022 -0.143 0.002
corn harvested area > median (dummy, by CZ)
-0.304 0.019 0.075 0.347 0.321 0.000
experienced corn producer (dummy)
0.323 0.002 0.082 0.251 0.119 0.005
profit in CP (1 year lag) 0.504 0.000 0.393 0.000 0.284 0.000 VAT support in CP (1 year lag)
0.271 0.329 0.496 0.000 -0.305 0.013
Constant 1.786 0.000 3.165 0.000 1.046 0.000
Nr. of observations 1593
4687
6752
Prob > F 0.000
0.000
0.000
R-squared 0.520
0.448
0.573
Adjusted R-squared 0.513
0.446
0.572
Source: own calculations
Note: To keep the table at reasonable length, the time dummies considered in the
model are not reported. Arable land median: CZ1 – 1541.5 ha, CZ2 – 1822 ha, CZ3 –
2182 ha. Corn harvested area median: CZ1 – 168 ha, CZ2 – 271 ha, CZ3 – 172 ha. All
the monetary values are deflated and calculated per hectare of harvested area.
Sunflower. Compared to the previous two crops, there are no significant differences
regarding sunflower yield determinants. The major production region is climatic zone 3
and the results show that additional specialization towards sunflower [share of
sunflower in sowing area] has a negative effect in this production region as well as in
climatic zone 2. However, the yield level is positively affected almost by any
intensification measure (including fertilizer [fertilizers], labor [labor costs in sunflower
production], 3rd party services [3rd party services] or other material costs [other
material costs]). Similar to corn (but not wheat), seeds [seeds] have a positive effect.
Sunflower yield shows higher sensitivity to land quality [land rent per ha] in climatic
zone 3. Also larger farms [arable land > median (dummy, by CZ)] have higher yields in
climatic zone 3, but farms that harvest more area under sunflower [sunflower harvested
area > median (dummy, by CZ)] realize higher yields in climatic zone 2. Again we
obtain the positive relation between profitability and yield level [profit in CP per ha (1
year lag)] and the learning effect [experienced sunflower producer (dummy)] that
positively influences yield across all climatic zones. As in the case of corn, we observe a
positive effect of VAT reimbursement [VAT support in CP per ha (1 year lag)] only in
climatic zone 2.
31
Table 8. Parameter estimates of regression model of sunflower yield
determinants, balanced panel, 2008-2013
Climatic zone 1
Climatic zone 2 Climatic zone 3
Coef. P>|t| Coef. P>|t| Coef. P>|t|
harvested area (sunflower)
0.000 0.526 0.000 0.800 0.000 0.014
land rent per ha 0.175 0.237 0.109 0.008 0.276 0.000
labor costs in sunflower production
0.205 0.301 0.278 0.000 0.370 0.000
seeds 0.727 0.000 0.671 0.000 0.930 0.000
fertilizers 0.274 0.001 0.465 0.000 0.508 0.000
3rd party services 0.351 0.000 0.284 0.000 0.336 0.000
other material costs 0.461 0.000 0.285 0.000 0.394 0.000
depreciation 0.182 0.045 0.292 0.000 0.304 0.000
other production costs 0.270 0.000 0.270 0.000 0.314 0.000
share of sunflower in sowing area
0.459 0.179 -0.469 0.000 -0.160 0.000
animal production (dummy)
-0.104 0.107 -0.004 0.877 0.082 0.000
share of AP in TP 0.542 0.001 0.230 0.000 0.031 0.502
share of niche crops in CP -0.078 0.623 -0.380 0.000 -0.152 0.002
holding dummy -0.065 0.371 -0.045 0.161 0.035 0.157
arable land > median (dummy, by CZ)
0.061 0.333 0.033 0.163 0.035 0.015
sunflower harvested area > median (dummy, by CZ)
0.024 0.706 0.061 0.011 -0.017 0.207
experienced sunflower producer (dummy)
0.181 0.001 0.115 0.000 0.093 0.000
profit in CP (1 year lag) 0.237 0.000 0.148 0.000 0.189 0.000 VAT support in CP (1 year lag)
0.224 0.254 0.081 0.084 -0.051 0.142
Constant 0.795 0.000 1.245 0.000 0.627 0.000
Nr. of observations 688
4946
12185
Prob > F 0.000
0.000
0.000
R-squared 0.565
0.512
0.535
Adjusted R-squared 0.550
0.510
0.534
Source: own calculations
Note: To keep the table at reasonable length, the time dummies considered in the
model are not reported. Arable land median: CZ1 – 1931 ha, CZ2 – 1702 ha, CZ3 –
1888.5 ha. Sunflower harvested area median: CZ1 – 120 ha, CZ2 – 220 ha, CZ3 – 450
ha. All the monetary values are deflated and calculated per hectare of harvested area.
32
Estimation of technical efficiency in crop production
The importance of determinants influencing technical efficiency of Ukrainian crop
production differs across climatic zones. For instance, in climatic zone 1, labor [labor
costs in CP], land quality [land rent], and fertilizers [fertilizers] are among the most
important drivers while in region 2 labor [labor costs in CP], seeds [seeds], and
fertilizers [fertilizers] contribute the most. For climatic zone 3, we identify land quality
[land rent], other agricultural production used as inputs [other agricultural products],
electricity [electricity], fuels [fuels], and property rent [property rent] as most
important. The crucial role of labor costs on technical efficiency can be explained from
different perspectives. For instance, this effect might be caused by the competition for
and higher costs of skilled labor (e.g., operators of new machinery) or higher salaries
may indicate a wage system that appreciates productive employees.
Diversification of production, assessed by the dummy variable “animal production
(dummy)”, shows a positive effect on technical efficiency of crop production in climatic
zones 2 and 3. However, the share of animal production in total production [share of AP
in TP] negatively influences efficiency of crop production in climatic zones 1 and 2. One
possible explanation is the use of otherwise idle capacities (e.g., buildings) for animal
production to cover some of the fixed costs. Increasing shares of animal production,
however, might cause conflicts in the use of farm resources for crop production.
Furthermore, the share of niche crops in the production structure [share of niche crops
in CP] has no statistically significant influence on technical efficiency. Regarding the size
of the farm or its integration into an agroholding, results are mostly ambiguous. Larger
farms [arable land > 1131 ha (median, dummy)] tend to have slightly higher TE in
climatic zones 1 and 3, while an agroholding-membership [holding dummy] impacts TE
negatively in the same regions.
Farms with higher profit per hectare [profit in CP per ha (1 year lag)] have usually also
higher technical efficiency and this effect is comparably high across all regions. VAT
reimbursement [VAT support in CP per ha (1 year lag)] does not show a statistically
significant effect on TE in climatic zones 1 and 2 but there is a negative effect in
climatic zone 3.
33
Table 9. Parameter estimates of regression model of technical efficiency
determinants, balanced panel, 2008-2013
Climatic zone 1
Climatic zone 2 Climatic zone 3
Coef. P>|z| Coef. P>|z| Coef. P>|z|
arable land 0.000 0.007 0.000 0.000 0.000 0.161
land rent 0.097 0.000 0.061 0.000 0.108 0.000
labor costs in CP 0.114 0.000 0.071 0.000 0.027 0.012
seeds 0.034 0.033 0.065 0.000 0.044 0.000
other agricultural products
0.012 0.724 0.064 0.003 0.105 0.018
fertilizers 0.086 0.000 0.072 0.000 0.058 0.000
oils -0.087 0.000 -0.034 0.001 -0.046 0.000
electricity 0.000 0.995 -0.008 0.887 0.071 0.036
fuels 0.077 0.134 0.026 0.346 0.073 0.000
spare parts 0.016 0.305 0.027 0.001 0.014 0.086
services in CP and other material costs
0.038 0.000 0.045 0.000 0.052 0.000
property rent -0.249 0.076 -0.137 0.052 0.191 0.055
depreciation in CP 0.013 0.275 -0.001 0.865 -0.028 0.000
animal production (dummy)
0.006 0.440 0.010 0.027 0.009 0.001
share of AP in TP -0.034 0.038 -0.042 0.000 0.010 0.312
share of niche crops in CP 0.010 0.445 -0.007 0.373 0.015 0.141
holding dummy -0.033 0.001 -0.004 0.462 -0.010 0.057
arable land > 1131 ha (median, dummy)
0.014 0.028 0.005 0.222 0.010 0.000
profit in CP (1 year lag) 0.045 0.000 0.039 0.000 0.044 0.000 VAT support in CP (1 year lag)
0.024 0.221 -0.008 0.341 -0.015 0.051
Constant 0.317 0.000 0.336 0.000 0.275 0.000
Nr. of observations 2693
6329
12590
Log likelihood 1530.53
7 3928.59
0 8066.61
6
Wald chi2 (24) 1110.54
0 2044.36
0 2660.96
0
Prob > chi2 0.000
0.000
0.000
Source: own calculations
34
3.3. Comparing differences among farms of different profitability levels
with similar structural characteristics
In this section we provide results of the treatment effect analysis (TEA) where we
compare developments of TFP, TE, growth indicators, and the input structure of farms
based on their profitability in the base year (2008). Farms with a profitability level
above 13.9% (the median of all farms) are coded as “1” and belong to the treated set,
while farms with lower profitability are in the non-treated group (coded as “0”).
The treated group shows on average higher TFP change than less profitable farms with
similar structural characteristics, but their TE change is significantly lower (see Table
10). We also observed no significant differences in TFP and TE changes comparing
farms with different land endowments (based on median of 2008).
Besides the base year (2008), we conducted this analysis also for the final year of the
dataset (2013) to investigate the dynamics of the results. It showed, that farms with
higher profitability in 2008 featured higher TE scores (on average by 0.133) but due to
moderate growth rates the difference significantly decreased until 2013.
Table 10. Treatment effect analysis of TFP and TE changes in farms with high
and low profitability levels in base year, 2008-2013
Dependent variable Target
population Number of
observations Coefficient P>|z|
TFP change* All farms 4497 0.286 0.000 TE in 2008* All farms 4497 0.133 0.000 TE change* All farms 4497 -0.330 0.000 TE in 2013* All farms 4497 0.021 0.000
Source: Own calculations
* without exact matches on regions
Figure 14 also illustrates the relations between base-year TE scores and TE change over
the period of investigation. Thus, more profitable farms, which feature higher technical
efficiency, show less TE change due to decreasing marginal returns.
35
Figure 14. Correlation of TE in base year and TE change
Source: own calculation
The application of TEA to growth indicators shows that farms with above median
profitability (in 2008) have lower growth of crop production but the significantly higher
production level (on average by 1537.3 tsd UAH7) compared to farms with below
median profitability. Between the two groups, there are no significant differences in
arable land and labor, while the treated group outperforms the control group in terms
of profit per hectare and crop yield. Again, growth rates for farms of the treated group
are lower but the levels of profit and yield per ha are still higher in 2013 compared to
the non-treated group.
Table 11. Treatment effect analysis of growth indicators in farms with high
and low profitability levels in base year, 2008-2013
Dependent variable Target
population Number of
observations Coefficient P>|z|
CP value in 2008* All farms 4497 1537.538 0.000 CP value – absolute
growth* All farms 4497 -1398.799 0.000
CP value – relative growth*
All farms 4497 -0.495 0.000
7 Median CP volume in 2008 – 3901.6 tsd UAH
01
23
45
67
TE
ch
an
ge
20
08
-20
13
(B
P)
0 .2 .4 .6 .8 1TE 2008 (BP)
TE change 2008-2013 (BP) Fitted values
36
Dependent variable Target
population Number of
observations Coefficient P>|z|
CP value in 2013* All farms 4497 138.739 0.659
Arable land in 2008* All farms 4497 87.119 0.146 Arable land – absolute
growth* All farms 4497 -55.371 0.335
Arable land – relative growth*
All farms 4497 -0.021 0.715
Arable land in 2013* All farms 4497 31.747 0.643
Labor in CP in 2008* All farms 4497 1.089 0.491 Labor in CP – absolute
growth* All farms 4497 -0.494 0.698
Labor in CP – relative growth*
All farms 4497 0.002 0.938
Labor in CP in 2013* All farms 4497 0.594 0.665
Source: Own calculations
* without exact matches on regions
Table 12. Treatment effect analysis of performance indicators in farms with
high and low profitability levels in base year, 2008-2013
Dependent variable Target
population Number of
observations Coefficient P>|z|
Profit per ha in CP in 2008*
All farms 4497 0.513 0.000
Profit per ha in CP – absolute growth*
All farms 4497 -0.217 0.000
Profit per ha in CP – relative growth*
All farms 4491 -1.315 0.621
Profit per ha in CP in 2013*
All farms 4497 0.296 0.000
Crop yield in 2008* All farms 4497 0.415 0.000 Crop yield – absolute
growth* All farms 4497 -0.230 0.001
Crop yield – relative growth*
All farms 4497 -0.354 0.000
Crop yield in 2013* All farms 4497 0.185 0.006
Source: Own calculations
* without exact matches on regions
On the input side, we see that higher profitability is correlated with lower costs. In
2008, more profitable farms had paid less for rented land, on average 23UAH/ha
(approximately 13% below the median) compared to less profitable farms.
Differentiation with respect to the investigated regions shows that there were no
significant differences in the first climatic zone, while the rental price for farms in the
37
treated group was 30(21) UAH/ha lower relative to the non-treated group in climatic
zone 2(3). Because of a less dynamic price development in the treated group, this gap
increases over the years, i.e., farms with higher profitability pay less land rent and the
price increase is smaller relative to the non-treated group. Accordingly, the price
difference in 2013 increased to 43 UAH/ha (ca. 7% of the median, see Table 13).
Material costs of the treated group were lower on average by 34 UAH (ca. 3% of the
median) per hectare in 2008 but in 2013 the difference between both groups was not
different from zero.
Farms with higher profitability seem to use superior (modern) technology (indicated by
higher capital assets), while less profitable farms rather rely on third-party services and
have a tendency to increase their use. Additionally, the treated group has higher labor
productivity (lower labor costs per hectare in CP) both in the base and the final year.
The observable increase in labor costs is higher for more profitable farms but mostly
reflects an increase of salaries. These results indicate that substitution of labor by
capital (i.e., mechanization and modernization) plays a major role in the treated relative
to the untreated group of farms.
Table 13. Treatment effect analysis of inputs in farms with high and low
profitability levels in base year, 2008-2013
Dependent variable Target
population Number of
observations Coefficient P>|z|
Land rent per ha in 2008*
All farms 4301 -0.023 0.000
Land rent per ha – absolute growth*
All farms 4236 -0.019 0.039
Land rent per ha – relative growth*
All farms 4236 -0.411 0.162
Land rent per ha in 2013*
All farms 4336 -0.043 0.000
Material costs in CP per ha in 2008*
All farms 4497 -0.034 0.000
Material costs in CP per ha – absolute growth*
All farms 4497 0.043 0.277
Material costs in CP per ha – relative growth*
All farms 4496 0.030 0.551
Material costs in CP per ha in 2013*
All farms 4497 0.010 0.812
Material costs in CP per ha – absolute growth
Farms of the climatic zone №1
560 0.326 0.016
Material costs in CP per ha – absolute growth
Farms of the climatic zone №2
1325 -0.188 0.037
38
Dependent variable Target
population Number of
observations Coefficient P>|z|
Material costs in CP per ha – absolute growth
Farms of the climatic zone №3
2612 0.077 0.068
Depreciation in CP per ha in 2008*
All farms 4163 0.011 0.009
Depreciation in CP per ha – absolute growth*
All farms 4085 0.040 0.000
Depreciation in CP per ha – relative growth*
All farms 4085 0.702 0.423
Depreciation in CP per ha in 2013*
All farms 4288 0.055 0.000
Third-party services in CP per ha in 2008*
All farms 4497 -0.022 0.000
Third-party services in CP per ha – absolute
growth*
All farms 4337 -0.034 0.069
Third-party services in CP per ha – relative growth*
All farms 3895 0.081 0.974
Third-party services in CP per ha in 2013*
All farms 4337 -0.047 0.005
Labor costs in CP per ha in 2008*
All farms 4497 -0.033 0.000
Labor costs in CP per ha – absolute growth*
All farms 4497 0.012 0.097
Labor costs in CP per ha – relative growth*
All farms 4492 0.092 0.469
Labor costs in CP per ha in 2013*
All farms 4497 -0.021 0.009
Source: Own calculations
* without exact matches on regions
3.4. Summary and discussion of the results
On average, Ukrainian farms feature low technical efficiency, which highlights
considerable farm heterogeneity in terms of production performance. Based on our
differentiation of production regions, we find that water restricted regions (as climatic
zone 2 and 3) are especially prone to inefficiencies. This highlights the importance of
water in particular and weather conditions in general regarding the efficiency of farms.
While adverse weather events (as drought or very cold/long winter) might have
negative effects for all farms, local conditions and/or managerial potential play an
important role by dealing with such circumstances (Deininger et al., 2015).
39
Despite the observed high fluctuations in its development, there is a positive trend in
farm productivity, especially manifested in the gradually increasing yield levels (of major
crops). However, not all (or even not the majority of) farms can benefit from such
development which also means that a relatively small share of farms is mainly
responsible for efficiency gains. As shown by Balmann et al. (2013) or Deininger et al.
(2015), one group of farms contributing to this trend are agroholdings – seemingly
answering the call by Zynich and Odening, (2009) to vertically and horizontally integrate
to better manage credit risk. However, we could not identify a similar restriction to one
farm organization. But more broadly, other - or the majority of – farms is lagging
behind the more productive ones, which gives rise to increasing inequality in farm
performances. The relevance of this finding is underlined by the positive correlation of
efficiency and farm profitability but also by the fact that there are 15 to 30 percent of
farms annually featuring negative profits. In this respect, it is not surprising that the
developments on the world market for agricultural products have major impacts on
productivity because Ukrainian farms depend highly on cash crops that are mostly
intended for export.
Our main focus is on crop yield levels and the analysis shows that there is considerable
potential to increase crop yield by intensification. Particular importance can be
attributed to inputs as fertilizer or seed – production components that can have a
considerable impact on (cost) efficiency and therefore require appropriate technology
and managerial knowledge. Despite the already large farm structure in our sample, we
identified positive size effects while no significant effect could be found regarding the
organization of farms, i.e., whether a farm belongs to a holding or not.
Land quality has a significant effect on the productivity of farms. Given the overall low
efficiency, this might indicate that (some) production systems are poorly adopted to the
local production conditions and instead aim to maximize short term benefits by the
engagement in cash crops such as wheat, corn, and sunflower. This is also supported
by the observation that in some regions the diversification of the crop rotation positively
affects efficiency, which hints the presence of phytosanitary problems in narrow
production systems.
The major finding of the presented analysis is that almost any indicator reacts positive
towards intensification. On the one hand, farms operating on low input levels aim to
minimize costs. On the other hand, if there are positive marginal profits by
intensification, there must be reasons to not extend input use. In fact, one might argue
that most of these reasons are exogenous to farms. For instance, one source of
inefficient production is limited access to required capital, i.e., financial means for short
or long term investments. This conclusion is also supported by empirical evidence of
imperfect capital markets (Zynich and Odening, 2009).
40
A second source for (persistent) production inefficiency on the input side of the farm
might relate to land market imperfections. Not only can the scattered landownership
and large scale-farm structure give rise for local market power and restrict the access to
the major production factor (Vranken and Swinnen, 2006), but the unresolved issue to
open and liberalize the land market is a prime source for inefficient resource allocation
(Deininger et al., 2015; Lioubimtseva et al., 2013). Land is one of the most important
production factors of agriculture and if its free transfer, e.g., from inefficient to efficient
farms is restricted, inefficiencies are conserved. Our results show (at least for some
regions) positive effects of farm growth on efficiency and productivity, which requires a
sufficiently dynamic structural change.
From a farm output perspective, further reasons for inefficiencies are the presence and
increasing exposure of agriculture to different types of risk. In general, if farms are risk
averse they will reduce input levels in the face of increasing uncertainty (Chavas, 2006).
For instance, Skakun et al. (2015) show that drought risk is non-uniformly distributed in
geographical space, can be substantial, and requires for an appropriate insurance
system, which is often not available (Shynkarenko, 2007). As a consequence, farms at
different locations in a region face and are affected by risk differently without adequate
measures to manage it. Zynich and Odening (2009) also highlight the need to reduce
credit risk by appropriate risk management instruments, while farms also have to deal
with considerable and increasing market risk in terms of price volatility (Bellemare et
al., 2015). Those and similar sources of risk may cause low input levels and inefficiency
in production by Ukrainian farms.
Furthermore, the dependence to export agricultural raw products to be processed
abroad might relate to capacity constraints of domestic processing. On the one hand,
processing itself has to consider domestic demand for processed agricultural products
and if this demand is weak, the processing sector has little or no incentive to invest.
Accordingly, excess production of agricultural raw products (i.e., beyond the level of
domestic processing in downstream markets) will need to go abroad. This, however,
will transfers value added out of Ukraine. On the other hand, production for the
international market often suffers from high regulation and risk exposure for Ukraine
processors. In fact, only the sunflower processing is well developed in the sense that
more than 90% of sunflower seed are processed into sunflower oil and almost all of it is
exported each year. The contrary is the case for soya and rapeseed. In fact, there are
initiatives concerning this issue but, e g., the attempt to implement a biofuel sector
based on rapeseed to participate in this dynamic market, were unsuccessful
(Schaffartzik, 2014).
Governmental instruments as subsidies can have a positive effect on farm productivity
(Zhu and Lansink, 2010). A recent study by Curtiss et al. (2017) shows that these
41
effects vary across different farm types depending on farm size or organization, and the
time period (subsidy level). In fact, subsidies might have the potential to decrease
(increase) average farm efficiency if already efficient (inefficient) farms benefit over-
proportional from those transfers. Given the low level of government payments after
2008, such subsidy effects are rather low if not negligible. The regression in our
analysis accounts for a specific type of subsidy – VAT reimbursement. In contrast to
direct subsidies, this support mechanism became a major source of agricultural
production subsidization. Despite the absent (climatic zones 1 and 2) or negative
(climatic zone 3) effect on technical efficiency, we found a positive influence of these
payments on yields, In this case though, the impact is heterogeneous across different
climatic zones and crops. Furthermore, we need to note that most support mechanism
themselves usually induce distortions by incentivizing or preserving inefficient market
allocations of resources. The implementation of any policy instruments therefore need
to focus on the mitigation of the aforementioned market imperfections on capital and
land markets and the development of the insurance system. Appropriate measures can
help to enable or support farms to benefit from risk reducing options as insurance or
irrigation systems, which offers the potential to increase farm productivity, e.g., crop
yields.
4. Concluding remarks
In aiming to answer the question whether Ukrainian farms are able to considerably
increase total crop production, we identify substantial inefficiencies. While this might
provide indeed the potential for significant productivity improvements, some of the
necessary conditions to access this potential are exogenous to Ukrainian farms.
Whether the supposedly large yield gap can be reduced, will therefore hinge on
numerous external factors in the realm of the institutional, economical, and social
framework in Ukraine as well as the natural conditions. As always, simple or general
solutions are not available but some basic conclusions can be made. For instance,
imperfections on input and output markets restrict farm development towards more
efficient production outcomes. While there are certainly considerable sources of
inefficiencies within the farm sector, (better) functioning markets for land and capital as
well as on the farm output side may provide major incentives to reallocate resources
and boost farm efficiency and productivity. On the other hand, there might be adverse
conditions as climate change that will affect the future prospect of crop production and
even might decrease yield levels (Mueller et al. 2016).
42
References
AgriSurvey (2013): Largest Agroholdings of Ukraine. Kyiv: UCAB.
Balmann, A., Curtiss, J., Gagalyuk, T., Lapa, V., Bondarenko, A., Kataria, K., & Schaft,
F. (2013). Productivity and efficiency of Ukrainian agricultural enterprises. Agriculture
Policy Report. APD/APR/06.
Bulava L.M. (2008). Physical Geography of Ukraine. – Kharkiv: AN GRO plus, 2008. –
224.
Coelli, T. J., Rao, D. S. P., O'Donnell, C. J., & Battese, G. E. (2005). An introduction to
efficiency and productivity analysis. Springer Science & Business Media.
Curtiss J., Gagalyuk T., Ostapchuk I. (2017). Effect of Public Subsidies on Productivity
of Crop Producing Farms in Ukraine – A Farm-level Difference-in-differences Analysis.
Paper prepared for the XV EAAE Congress, August 29th – September 1st , 2017, Parma,
Italy.
Deininger, Klaus, Denys Nizalov, and Sudhir K. Singh. "Determinants of productivity and
structural change in a large commercial farm environment: Evidence from Ukraine." The
World Bank Economic Review (2015).
Kulyk I., Herzfeld T. and Nivievskyi O. (2014). Comparative assessment of Ukrainian
grain export policies. Agriculture Policy Report. Kyiv: APD.
Lioubimtseva, E., de Beurs, K. M., & Henebry, G. M. (2013). Grain production trends in
Russia, Ukraine, and Kazakhstan in the context of the global climate variability and
change. In Climate change and water resources (pp. 121-141). Springer Berlin
Heidelberg.
Lobell, D. B., Cassman, K. G., & Field, C. B. (2009). Crop yield gaps: their importance,
magnitudes, and causes. Annual review of environment and resources, 34, 179-204.
Marc F. Bellemare; Rising Food Prices, Food Price Volatility, and Social Unrest. Am J
Agric Econ 2015; 97 (1): 1-21.
Matyukha, A., Voigt, P. and Wolz, A. (2015). Agro-holdings in Russia, Ukraine and
Kazakhstan: temporary phenomenon or permanent business form? Farm-level evidence
from Moscow and Belgorod regions. Post-Communist Economies, 27 (3): 370-394.
Müller, D., Jungandreas, A., Koch, F., & Schierhorn, F. (2016). Impact of Climate
Change on Wheat Production in Ukraine. Agricultural Policy Report APD/APR/02/2016.
German-Ukraine Agricultural Policy Dialogue.
43
Neumann, K., Verburg, P. H., Stehfest, E., & Müller, C. (2010). The yield gap of global
grain production: A spatial analysis. Agricultural systems, 103(5), 316-326.
Osborne, S. and Trueblood, M. A. (2002). Agricultural Productivity and Efficiency in
Russia and Ukraine. Agricultural Economics Report No 813. Washington, DC: Economic
Research Service, United States Department of Agriculture (USDA).
Ryabchenko, O., & Nonhebel, S. (2016). Assessing wheat production futures in the
Ukraine. Outlook on Agriculture, 45(3), 165-172.
Schaffartzik, A., Plank, C., & Brad, A. (2014). Ukraine and the great biofuel potential? A
political material flow analysis. Ecological Economics, 104, 12-21.
Schierhorn, F., Faramarzi, M., Prishchepov, A. V., Koch, F. J., & Müller, D. (2014).
Quantifying yield gaps in wheat production in Russia. Environmental Research Letters,
9(8), 084017.
Shynkarenko, R. (2007). Introduction of Weather Index Insurance in Ukraine-Obstacles
and Opportunities. In 101st Seminar, July 5-6, 2007, Berlin Germany (No. 9244).
European Association of Agricultural Economists.
Skakun, S., Kussul, N., Shelestov, A., & Kussul, O. (2016). The use of satellite data for
agriculture drought risk quantification in Ukraine. Geomatics, Natural Hazards and Risk,
7(3), 901-917.
SSSU (2013): Sowing areas of agricultural crops in 2013. State Statistic Service of
Ukraine. Statistical bulletin. Kyiv.
SSSU (2014a): Agriculture of Ukraine in 2013. State Statistic Service of Ukraine.
Statistical bulletin. Kyiv.
SSSU (2014b): Ukraine’s Foreign Trade. State Statistic Service of Ukraine. Statistical
yearbook. Kyiv.
SSSU (2014c): Main economic indicators of agricultural production in agricultural
enterprises in 2013. State Statistic Service of Ukraine. Statistical bulletin. Kyiv.
SSSU (2016): Agriculture of Ukraine in 2015. State Statistic Service of Ukraine.
Statistical bulletin. Kyiv.
StataCorp. (2015): Stata: Release 14. Statistical Software. College Station, TX:
StataCorp LP.
Troyer A. Forrest (1996). Breeding widely adapted, popular maize hybrids. Euphytica 92
(1-2), 163-174.
44
Vranken, L., Swinnen, F. (2006): Land rental markets intransition: Theory and evidence
from Hungary. World Development 34(3): 481-500.
Zinych, Nataliya, and Martin Odening. "Capital market imperfections in economic
transition: empirical evidence from Ukrainian agriculture." Agricultural Economics 40.6
(2009): 677-689.
Zhu, X. and A. Oude Lansink (2010). Impact of CAP subsidies on technical efficiency of
crop farms in
Germany, the Netherlands and Sweden. Journal of Agricultural Economics, 61 (3): 545-
564.
45
Appendix A. Summary statistics of variables for DEA model, 2008-2013
Variables Obs Mean Std. Dev. Min Max
DEA model variables - unbalanced panel
Climatic zone 1
CP value (tsd UAH) 7370 5443 13867 14 295482
Labor units in CP (persons) 7370 32 56 1 1309
Total land (ha) 7370 1636 2895 6 90621
Material costs and depreciations in CP (tsd UAH) 7370 3957 11575 4 424671
From that – depreciation 7370 400 1257 0 30322
- material cost (without services) 7370 3557 10768 2 418044
- services 7370 947 4551 0 233980
Climatic zone 2
CP value (tsd UAH) 13276 8343 25110 17 824351
Labor units in CP (persons) 13276 50 128 1 4161
Total land (ha) 13276 2365 5601 4 178083
Material costs and depreciations in CP (tsd UAH) 13276 5231 15603 8 450777
From that – depreciation 13276 547 1721 0 76789
- material cost (without services) 13276 4684 14432 8 447163
- services 13276 1143 4299 0 127464
Climatic zone 3
CP value (tsd UAH) 24255 5314 10434 19 370905
Labor units in CP (persons) 24255 38 80 1 4213
Total land (ha) 24255 2210 3630 8 136400
Material costs and depreciations in CP (tsd UAH) 24255 3228 6253 6 222737
From that – depreciation 24255 451 1039 0 54395
- material cost (without services) 24255 2777 5427 4 216445
- services 24255 574 1558 0 64274
DEA model variables - balanced panel
Climatic zone 1
CP value (tsd UAH) 3360 7056 14303 59 237306
Labor units in CP (persons) 3360 42 59 1 1217
Total land (ha) 3360 2023 2847 40 40503
Material costs and depreciations in CP (tsd UAH) 3360 4915 11283 30 268813
From that – depreciation 3360 528 1270 0 21812
- material cost (without services) 3360 4387 10436 27 266681
- services 3360 1099 4313 0 119675
Climatic zone 2
CP value (tsd UAH) 7949 9188 24593 99 603044
Labor units in CP (persons) 7949 56 129 1 3265
Total land (ha) 7949 2558 5625 81 178083
Material costs and depreciations in CP (tsd UAH) 7949 5654 14954 57 325126
From that – depreciation 7949 619 1890 0 76789
46
Variables Obs Mean Std. Dev. Min Max
- material cost (without services) 7949 5035 13623 40 319700
- services 7949 1139 3614 0 126824
Climatic zone 3
CP value (tsd UAH) 15673 6320 11776 76 370905
Labor units in CP (persons) 15673 45 92 1 4213
Total land (ha) 15673 2578 4177 60 136400
Material costs and depreciations in CP (tsd UAH) 15673 3781 6993 21 222737
From that – depreciation 15673 547 1162 0 54395
- material cost (without services) 15673 3235 6066 21 216445
- services 15673 629 1680 0 64274
Note: Monetary values are expressed in nominal values.
Source: Own calculations
Appendix B. Summary statistics of variables for DEA model for individual years, 2008-2013
2008 2009 2010 2011 2012 2013
Variables Obs Mean Std. Dev.
Obs Mean Std. Dev.
Obs Mean Std. Dev.
Obs Mean Std. Dev.
Obs Mean Std. Dev.
Obs Mean Std. Dev.
DEA model variables - unbalanced panel
Climatic zone 1 CP value (tsd UAH)
1462 3858 7705 1347 3919 9057 1156 4335 9984 1196 5845 16067 1119 7440 17266 1090 8140 20150
Labor units in CP (persons)
1462 33 43 1347 30 42 1156 33 53 1196 31 55 1119 33 63 1090 35 79
Total land (ha)
1462 1314 1797 1347 1417 2078 1156 1641 2510 1196 1678 3078 1119 1865 3213 1090 2053 4351
Material costs and depreciations in CP (tsd UAH)
1462 2447 5070 1347 2450 6243 1156 3152 8018 1196 4216 11442 1119 5671 14816 1090 6655 19230
From that - depreciation
1462 172 728 1347 299 791 1156 416 1179 1196 422 1044 1119 562 1599 1090 626 1937
- material cost (without services)
1462 2275 4640 1347 2152 5694 1156 2736 7231 1196 3794 10818 1119 5109 13725 1090 6029 18010
- services 1462 442 1477 1347 514 2391 1156 710 2919 1196 1071 4676 1119 1425 4975 1090 1787 8376
Climatic zone 2 CP value (tsd UAH)
2314 6928 18335 2248 7208 22603 2170 6949 21787 2166 8866 27424 2185 9718 26675 2193 10493 31726
Labor units in CP (persons)
2314 53 135 2248 48 103 2170 51 132 2166 49 129 2185 51 148 2193 49 119
Total land (ha)
2314 2199 5198 2248 2258 4922 2170 2427 6169 2166 2365 5597 2185 2458 5740 2193 2496 5928
Material costs and depreciations in CP (tsd UAH)
2314 4152 12152 2248 3965 11119 2170 4368 13647 2166 5287 15680 2185 6440 18119 2193 7264 20696
From that - depreciation
2314 259 723 2248 496 2039 2170 503 1265 2166 595 1639 2185 700 1952 2193 749 2221
- material cost (without services)
2314 3892 11605
2248 3468 9687 2170
3865 12757
2166 4692 14461 2185 5741 16688 2193 6515 19254
48
2008 2009 2010 2011 2012 2013
Variables Obs Mean Std. Dev.
Obs Mean Std. Dev.
Obs Mean Std. Dev.
Obs Mean Std. Dev.
Obs Mean Std. Dev.
Obs Mean Std. Dev.
- services 2314 867 3747 2248 811 2894 2170 979 4079 2166 1222 4846 2185 1415 4398 2193 1587 5392
Climatic zone 3 CP value (tsd UAH)
3883 5419 10123 3970 4881 9252 4115 4587 8598 4247 5563 10907 4022 5236 11116 4018 6201 12124
Labor units in CP (persons)
3883 44 99 3970 40 89 4115 37 82 4247 36 77 4022 35 63 4018 33 64
Total land (ha)
3883 2314 3692 3970 2279 3708 4115 2199 3602 4247 2140 3631 4022 2170 3573 4018 2166 3575
Material costs and depreciations in CP (tsd UAH)
3883 2965 5896 3970 2683 5636 4115 2650 5204 4247 3198 5955 4022 3677 7101 4018 4195 7318
From that - depreciation
3883 255 612 3970 431 974 4115 427 965 4247 477 1004 4022 544 1279 4018 562 1226
- material cost (without services)
3883 2709 5372 3970 2252 4811 4115 2222 4388 4247 2721 5141 4022 3132 6174 4018 3633 6309
- services 3883 504 1418 3970 462 1389 4115 491 1327 4247 577 1306 4022 622 1925 4018 789 1847
DEA model variables - balanced panel
Climatic zone 1 CP value (tsd UAH)
560 6165 9766 560 6030 10819 560 5407 9198 560 7652 15885 560 8476 17410 560 8605 19169
Labor units in CP (persons)
560 45 50 560 42 50 560 41 54 560 42 63 560 42 71 560 40 64
Total land (ha)
560 1842 2105 560 1950 2418 560 2005 2626 560 2097 3117 560 2110 3204 560 2135 3391
Material costs and depreciations in CP (tsd UAH)
560 3756 6339 560 3708 7733 560 3920 8406 560 5309 12449 560 6333 15797 560 6465 13468
From that - depreciation
560 275 560 560 477 980 560 537 1195 560 580 1219 560 648 1545 560 651 1736
- material cost (without services)
560 3481 5892 560 3231 7137 560 3383 7730 560 4729 1169
9 560 5686 14845 560 5814 12029
- services 560 694 2086 560 825 3551 560 844 3274 560 1317 5843 560 1549 6086 560 1365 3504
49
2008 2009 2010 2011 2012 2013
Variables Obs Mean Std. Dev.
Obs Mean Std. Dev.
Obs Mean Std. Dev.
Obs Mean Std. Dev.
Obs Mean Std. Dev.
Obs Mean Std. Dev.
Climatic zone 2 CP value (tsd UAH)
1325 7996 19880 1325 8191 23900 1325 7402 19049 1325 9998 27851 1325 10524 26523 1325 11016 28486
Labor units in CP (persons)
1325 60 156 1325 54 118 1325 55 118 1325 56 127 1325 57 128 1325 56 122
Total land (ha)
1325 2446 6029 1325 2496 5513 1325 2555 5764 1325 2609 5546 1325 2617 5387 1325 2622 5496
Material costs and
depreciations in CP (tsd UAH)
1325 4551 11197 1325 4506 12175 1325 4528 11175 1325 5832 15287 1325 6890 17843 1325 7620 19569
From that - depreciation
1325 308 769 1325 592 2473 1325 537 1121 1325 668 1739 1325 763 1962 1325 847 2536
- material cost (without services)
1325 4242 10586 1325 3913 10410 1325 3991 10513 1325 5165 13932 1325 6126 16391 1325 6773 17701
- services 1325 892 3111 1325 868 2951 1325 954 2987 1325 1280 4575 1325 1312 3481 1325 1530 4204
Climatic zone 3 CP value (tsd
UAH) 2612 6256 11216 2612 5722 10418 2612 5599 9819 2612 6942 12347 2612 6181 12496 2612 7220 13812
Labor units in CP (persons)
2612 49 111 2612 46 101 2612 45 96 2612 45 92 2612 42 71 2612 41 75
Total land (ha)
2612 2587 4229 2612 2605 4274 2612 2608 4207 2612 2591 4157 2612 2560 4118 2612 2517 4075
Material costs and depreciations in CP (tsd UAH)
2612 3349 6610 2612 3100 6401 2612 3197 5986 2612 3923 6680 2612 4310 7974 2612 4809 7902
From that -depreciation
2612 296 694 2612 509 1079 2612 529 1089 2612 612 1116 2612 659 1450 2612 675 1350
- material cost (without services)
2612 3052 6005 2612 2591 5482 2612 2668 5063 2612 3311 5761 2612 3651 6958 2612 4134 6763
- services 2612 548 1597 2612 502 1577 2612 566 1526 2612 652 1388 2612 683 2121 2612 822 1756
Note: Monetary values are expressed in nominal values.
Source: Own calculations
______________________________________________________________ANNEX TO CHAPTER 6
50
Appendix C. Total factor productivity change, 2008-2013
Variable 2008~2009 2009~2010 2010~2011 2011~2012 2012~2013 Cumulative
Climatic zone 1
TFP change 1.060 1.108 0.894 0.990 1.028 1.069
Technical
change 0.954 1.181 0.812 1.285 0.846 0.995
Efficiency change
1.111 0.938 1.101 0.771 1.215 1.075
Pure efficiency
change 1.078 0.911 1.082 0.835 1.219 1.082
Scale efficiency
change 1.031 1.030 1.018 0.923 0.997 0.995
Climatic zone 2
TFP change 1.014 1.104 0.864 1.000 1.003 0.970
Technical
change 1.006 1.050 0.912 1.079 1.078 1.121
Efficiency
change 1.008 1.052 0.947 0.927 0.930 0.866
Pure efficiency change
0.999 1.066 0.925 0.922 0.964 0.876
Scale efficiency
change 1.009 0.986 1.024 1.005 0.965 0.988
Climatic zone 3
TFP change 1.019 1.039 0.940 1.207 0.901 1.082
Technical change
1.123 1.016 0.914 1.175 0.888 1.088
Efficiency
change 0.907 1.022 1.029 1.027 1.015 0.994
Pure efficiency
change 0.943 1.015 1.037 1.029 1.021 1.043
Scale efficiency change
0.962 1.007 0.992 0.999 0.994 0.954
Note: The values are geometric means of individual farm values.
Number of observations: CZ1 - 560, CZ2 - 1324, CZ3 - 2612 (numbers are equal for the same
climatic zone across years).